《牛津洞见:2023年全球政府人工智能就绪指数评估报告(英文版)(53页).pdf》由会员分享,可在线阅读,更多相关《牛津洞见:2023年全球政府人工智能就绪指数评估报告(英文版)(53页).pdf(53页珍藏版)》请在三个皮匠报告上搜索。
1、Government AIReadiness Index2023CreditsThisreportwasproducedbyOxfordInsights.Authors:EmmaHankinsPabloFuentesNettelLiviaMartinescuGonzaloGrauSulamaanRahimEditedby:SulamaanRahimThisversionwaspublishedon6December2023FThisworkislicensedunderaCreativeCommonsAttribution-ShareAlike4.0InternationalLicense.2
2、023OxfordInsights2GovernmentAIReadinessIndex2023Table of ContentsCredits2TableofContents3TheGovernmentAIReadinessIndex20234OurFramework5Globalfindingsinfull6RegionalReports12NorthAmerica13LatinAmerica&TheCaribbean16WesternEurope18EasternEurope22MiddleEast&NorthAfrica24Sub-SaharanAfrica26South&Centra
3、lAsia28EastAsia30Pacific32AnnexI:Methodology34AnnexII:Detailedscores433The Government AI Readiness Index 2023In 2023,artificial intelligence(AI)was in theheadlinesmorethanever.GenerativeAIbreakthroughs,major developments in the field of AI regulation like the European Unions AI Act,and a significant
4、increase in AI-related summits globallyhaveputthistechnologyinthespotlight.ThetransformativepotentialofAIisundeniable,withgovernmentsworldwideacknowledgingitsimpact.Governments are not only working to regulate AI and foster AI innovation,but also strivingtointegrate this technology into public servi
5、ces.Countries like the RepublicofKoreaareusingAItoimprove government operations through the Digital Platform Government.Similarly,the UKsNational Health Service is supporting the researchandinnovationofnewAIscreeningtechnologiesforhealthandsocialcare.However,understanding how to ensure that AI is ad
6、opted effectivelyforthepublicgoodremainsachallenge.This index attempts to address this challenge.Our primary research question remainsunchanged:how ready is a given government to implementAIinthedeliveryofpublicservicestotheircitizens?This year,we are expanding our scope to rank 193 countries,up fro
7、m 183 in the 2022 edition.Werecognise that government AI readiness is a global concern,and our goal is to include as manycountries as possible intheindexrankings.Thisguidestheselectionofourindicatorstoensuredataisavailableforthemajorityofcountries.This report highlights the main findings for each of
8、 our pillars and provides insights into globaltrendsintheAIgovernancelandscape.Additionally,asinpreviousyears,weincluderegionalreportsanalysing the major trends and initiativesaffectingeachregionsAIreadiness1.Ouranalysisisbasedon a combination of our index scores and complementary desk research and
9、analysis.Due to thecomplexity and breadth of the index,it is not always possible to draw a clearcausallinebetweenaparticular policy or event and a change in score in a specificindicator.Ourgoalhasbeentoprovidebroader insights into regional and national AI policy contexts than can be provided by nume
10、ricalscoresalone.1We have divided the world into 9 regions,based on a combination of the UN and the World Bankregionalgroupings.4GovernmentAIReadinessIndex2023Our FrameworkWeinclude39indicatorsacross10dimensions,whichmakeup3pillars:GovernmentA government should have a strategic vision for how it dev
11、elops and governs AI,supported byappropriate regulation and attention to ethical risks(governance and ethics).Moreover,it needs tohave stronginternaldigitalcapacity,includingtheskillsandpracticesthatsupportitsadaptabilityinthefaceofnewtechnologies.TechnologySectorA government depends on a good suppl
12、y of AI tools from the countrys technology sector,whichneeds to be mature enough to supply the government.The sector should have high innovationcapacity,underpinned by a business environment that supports entrepreneurship and a good flowof R&D spending.Good levels of human capital the skills and edu
13、cation ofthepeopleworkinginthissectorarealsocrucial.Data&InfrastructureAI tools need lots of high-quality data(data availability)which,to avoid bias and error,should alsobe representative of the citizens in a given country(data representativeness).Finally,this dataspotential cannot be realised witho
14、uttheinfrastructurenecessarytopowerAItoolsanddeliverthemtocitizens.Figure1:ThePillarsoftheGovernmentAIReadinessIndex5Global findings in fullGovernmentPillarThe number of AI strategies released per year is trending downward,butthepicturelooksmorediverse.The overall number of published AIstrategies ha
15、s decreased comparedto previous years.This decline isprimarily attributed to fewer strategiespublished by countries in the higherincomeend.2However,its worth noting that thisyear marks the most diverse collectionof new or upcoming AI strategies todate.Half of the AI strategies thatwere published or
16、announced comefrom low and lower middle incomecountries.This represents a significant change as in previous years AI workwasdominatedbyhighanduppermiddleincomecountries(seeFigure2).Notably,Rwanda has become the first country within the low income bracket to publish an AIstrategy.Similarly,lower midd
17、le income countries,including Tajikistan,Senegal,and Benin,havepublishedAIstrategiesthisyear,whileEthiopiaandSriLankaaresettoreleasetheirs.The picture is also more diverse in terms ofregions.Countries that have either announcedor published AI strategies this year are spreadacross Latin America&the C
18、aribbean,South&CentralAsia,Sub-Saharan Africa,and theMiddle East and North Africa.These regionshave historically been underrepresented in AIstrategy work,with less than 50%of theircountrieshavingreleasedastrategy.Additionally,theseregions currently scorebelow the global average in the Governmentpill
19、ar.It is therefore encouraging to observetheir efforts inestablishingrobustfoundationsforAIgovernance,indicatingapotentialforimprovedperformanceinthispillarinthefuture.2Our income group classification is based on the World Banks Group country classifications by income level for2023-2024.6GovernmentA
20、IReadinessIndex2023The release and announcement of strategies by countries in the lower-incomebracketshintata potential momentum boost among nations that were lagging in formalising their AI visions.AI,as the globalgovernancetrendssectionhighlights,isconcurrentlybecomingapivotalelementinthe internat
21、ional development agenda,with cooperation agencies sponsoring and advisingdeveloping countries on AI governance.These elements could drive a potential surge in the overalldevelopmentofAIstrategiesinthecomingyears.DataandInfrastructurePillarThedigitaldivideremainsaglobalchallenge.Scores in the Data a
22、nd Infrastructure pillarshow a substantial difference between highand low income economies,illustrating anexisting global digital divide.These differencesare particularly pronounced in theaveragescoresof low and high income groups within thedimensions of Data Availability andInfrastructure(seeFigure
23、4).The disparity in these two dimensions is alsoevident when looking across regions.TheInfrastructuredimensionreveals one of thehighest gaps between the highest and lowest-ranked regions,withNorthAmericascoring80.57andSub-Saharan Africa scoring 32.44(see Figure 5).Notably,there is a substantial disp
24、arity in DataAvailability,marked by an 18-point difference between the lowest-ranked,Sub-Saharan Africa(32.44),and the second-lowest-ranked,Pacific(50.79).Thisunderscoresapronouncedimbalance.When considering all dimensions,the average gap between the lowest andsecond-lowestregionsis6.10points,highli
25、ghting a significant challenge in terms ofdataavailabilityforSub-SaharanAfrica.This is particularly relevant as both dataavailabilityandinfrastructurearecriticalenablers for AI readiness.The development ofAI systems requires extensive volumes of datafor tasks such as training models,refiningalgorith
26、ms,and mitigating bias,among othercrucial aspects.At the same time,establishinga robust infrastructure foundation is essential7to facilitate the operability and scale of AItools,aswellastoguaranteeequitableandsafeaccesstothem.Amid the generative AI boom,which puts the spotlight on major risks like p
27、rivacy,labourdisplacement,and misinformation,its important not to underestimate the effects of theexisting global digital gap.While the emergence of generative AI models holds the potential forsignificantimprovementsinpublicservicesforcountriesinthelowerincomebracket,theassociatedrisks must be ackno
28、wledged.Without a solid base of data and infrastructure,countries may find itchallenging to develop domestic generativeAIcapabilities,potentiallyleadingtorelianceonforeigntechnology.This reliance could introduce additional hurdles,including the unavailabilityofAItoolsin local languages and the poten
29、tial for biases in AI models.Addressing these challenges becomesessentialforfosteringequitableandinclusiveadvancementsinAIreadiness.TechnologySectorPillarThere is a disparity between high income countriesandeveryoneelse,though some large middle income economies punch above theirweight.The data also
30、suggests adisparitywithintheTechnology Sector pillar.However,thisdisparityappearstobe betweenhigh income countries andcountries in every otherincome group.This is bestillustrated by looking attheaveragescoresforeachincome group within eachdimensionoftheTechnologySectorpillar(see Figure 6).Across all
31、three dimensions,the gap in average scores between low income and lower middle incomecountries is relatively small(3 to 8 points),as is the gap in average scores betweenlower-middle-income and upper-middle-income countries(2 to 7 points).However,the gap inaverage scores between upper-middle-income a
32、nd high-income countries is much larger at 13to 17 points in the Innovation Capacity and Human Capital dimensions,this disparity is greaterthan the gaps between every other income group combined.The US dominates the TechnologySectorpillarwithascore12pointshigherthanthesecond-rankingUnitedKingdomcomp
33、aredtoa2-point difference between first and second-ranking countries in the other pillars.However,even ifweexcludethisoutlier,thegapnarrowsonlyslightlyby0.27to0.80points.8GovernmentAIReadinessIndex2023That being said,it is important to note that some lower-income countries punch above theirweight,sc
34、oring closer to countries in the high income group.While high income countriesdominate the top of most indicators in the Technology Sector pillar,there are notable exceptions.Excluding imputed data,Malaysia is the non-high income country most frequently listed in the top20 countries in these indicat
35、ors,appearing as o?en as the high-income countries oftheRepublicofKorea,Australia,or Ireland.This aligns with Malaysias high overall score of 68.71,which is muchhigher than the average score of 43.69 for upper middle income countries and leads it to rank23rdglobally.Malaysias strength mostly lies in
36、 dimensions relatedtoAIskillsandeducation,performingwell inICTskills,GraduatesinSTEMorcomputerscience,Qualityofengineeringandtechnologyhighereducation,and AI researchpapers.Thissuggeststhecountryissetuptobeasourceofmuch-neededAItalentintheyearstocome.Malaysia is followed closely by the large non-hig
37、h income economies of China,India,Brazil,andthe Russian Federation sometimes called the BRIC countries and China ranks tenth in theworld in the Technology Sector pillar.Whatliesbehindthesescores?China,Brazil,andIndiaallranknear the topinNumberofAIunicornsandNumberofnon-AItechunicorns,likelybenefitin
38、gfromhugeconsumer populations capable of powering billion-dollar start-ups.All BRIC countries are alsorepresented in the top 20 for AI research papers(where India outranked the US for the first time in2023 to take 2nd place behind China)and Quality of engineering and technology higher education.This
39、 suggests that these countries large populations have access to quality educational andresearchinstitutions,whichiscriticalforfuturehumancapitalandresearchperformance.Even considering these outperforming middle-income countries,the largedisparityintechsectorreadiness is concerning.If a countrys dome
40、stic tech sector is too immature or lacking in humancapital or innovation capacity to create adequate AI tools,governments may be forced to turn toforeign companies,likely in higher-income countries,to procure AI services.This both stunts thegrowth ofthedomestictechsectorandcanhaveevenmoredireconseq
41、uencesforAI-enabledpublicservices,which may be improperly trained on foreign data not representative or relevant to acountryscontext.Globalgovernancetrends2023 saw increased international collaboration on AI,especially on AIgovernanceandethics.While the index focuses largely on individual countries
42、efforts on AI in public services,2023 hasbeen a very active year for global collaboration on AI.This is a welcome development,as theimpacts of AI cross national borders and AI supply chains are increasingly global.A robust globalgovernance framework for AI is therefore essential to equitably distrib
43、ute the benefits of thistechnologyandeffectivelyaddressandmitigateitsrisks.9We have seen a boom in AI summits,from the launch of the Hiroshima AI process during the G7Summit in May to the UKs AI Safety Summit and accompanying AI Fringe inNovember.Clearly,weare seeing increased awareness of thesociet
44、alimplicationsofAIandmajorpoliticalbodiesaretaking action.ThisyearhasseenthereleaseofseveralproposedwaysofgoverningAI,includingtheG7s International Guiding Principles for Advanced AI Systems and International Code of Conductfor Organizations Developing Advanced AI Systems,as well as the Bletchley De
45、claration thatemerged from the UK AI Safety Summit.These organisations and documents join others alreadyestablished in the field of AI governance,including the OECDs 2019 AI Principles,UNESCOs 2021RecommendationontheEthicsofArtificialIntelligence,andtheEuropeanUnionsdra?AIAct.This year has also seen
46、 an increase in AI governance collaboration at the regional level.InLatinAmerica and the Caribbean,twenty nations agreed to the Santiago Declaration a?er a summithosted by the Chilean government with the support of UNESCO and the Latin AmericanDevelopment Bank CAF.This follows the 2022 meeting of se
47、ven southern African countries inWindhoek,Namibia for a UNESCO-Southern Africa sub-Regional Forum on Artificial Intelligence,which agreed to the Windhoek Statement recommending actions on data,education,andgovernanceacrosstheregion.As the above summits suggest,AI has also become part of the internat
48、ional developmentagenda.We are now seeing AI governance developments beyond the Global North,inpartbecauseof support from cooperation agencies and development banks.These efforts are showing clearresults,with Rwanda and Senegal publishing national AI strategies this year with the support ofinternati
49、onal development organisations including GIZ FAIR Forward,the World Economic Forum,andtheAU-EUDigitalforDevelopment(D4D)Hub.UNESCOs activities in this area merit a special mention.The organisation has created twomethodologiestoenablecountriestoimplementtheirrecommendationsonAIethics:theReadinessAsse
50、ssment Methodology(RAM),which assesses whether a country is prepared to implement theUNESCO recommendations,and the Ethical Impact Assessment,which helps project teams assesspotential ethical impactsofAIsystemstheyaredeveloping.SupportedbytheGovernmentofJapan,the EuropeanCommission,andCAF,UNESCOhasa
51、nnouncedthatitwillworkwithaninitialgroupof50 countries to create country reports based on the RAM and tailor specific recommendations.Notably,these countries are largely developingcountriesintheGlobalSouth.Someofthecountriesthat have committed to working with UNESCO on implementing its ethics recomm
52、endations havenot yet published national AI strategies,meaning they may publish ethical AI guidelines beforeanyoverarching AI strategy.This would be a reversal of the process we have seen across almost everyother country and it has the potential to significantly change how countries approach AI ingo
53、vernment.10GovernmentAIReadinessIndex2023RegionalReportsGovernmentAIReadinessIndex202311North AmericaBy Emma HankinsTheregionallandscapeLooking at this years data,the story of the North American region appears to be one ofcontinuity:the United States and Canada continue to score near the top of the
54、index,withtheUSArankingfirst at 84.80 and Canada ranking fi?h at 77.07.The countries both have high scores in theGovernment and Data and Infrastructure pillars,but the US dramatically outperforms in theTechnology Sector pillar,scoring 12 points higher than anyothercountry.Beyondthenumbers,thepastyea
55、rhasbeenabusytimeforAIpolicydevelopmentsinbothcountries.KeydevelopmentsBoth the USA and Canada rank in the top five globally in the Government pillar,seeminglytrackingthe many AI governance initiatives that have been introduced in both countries recently.One of the most important developments in gov
56、ernment AI readiness in the region happened inOctober 2023,as US President Biden announced the Executive Order on the Safe,Secure,andTrustworthy Development and Use of Artificial Intelligence.Amongst other things,the sweepingorder directs every federal agency to appoint a Chief AI Officer and,for so
57、me agencies,to createinternal AI Governance Boards.Additionally,agencies are instructed to develop standards for thegovernments use of AI,according to the National Institute of Standards and Technology(NIST)sAI12GovernmentAIReadinessIndex2023Risk Management Framework,published January 2023.The execu
58、tive orderalsoincludesasectionon strengthening American leadership abroad and was released the same week as US VicePresident Kamala Harris spoke at the UK AI Safety Summit,suggesting the US intends to remainactiveinglobalAIgovernance.While the specific laws and regulations derived from this executiv
59、e order will be created byCongress,it does impose some mandates by invoking the 1950 Defense Production Act.It requirescompaniesdevelopinglarge-scaleAIsystemsthatcouldaffectnationalsecurity,publichealth,ortheeconomy to test these systems and report results to the government.It also orders rules to b
60、edra?ed around federal procurement of AI,which will have a large impact due to the governmentsroleasamajorpurchaserofadvancedtechnologylikeAI.The order also requires the Department of Commerce to develop guidance on labellingAI-generated content and requires federal agencies to watermark such conten
61、t.This rule and theorders wider callout of the risk of deepfakes and generative AI is particularly timely ahead of theNovember 2024 general election in the US.Much of thecampaignwillplayoutonsocialmedia,andexperts have expressed serious concern about the potential for AI products such as deepfakes t
62、ofosterwidespreaddis-andmisinformation.Canada,meanwhile,was one of the first countries to respondofficiallytothegenerativeAIboomofearly 2023,publishing its Guide on the use of Generative AI for federal institutions in September2023.This guide outlines the challenges that generative AI poses,as well
63、as outlining a number ofbest practices and additional resources for federal institutions to draw upon when using the newtechnology.At the same time,Canada also published a Voluntary Code of Conduct on theResponsible Development and Management of Advanced Generative AI Systems aimed at privatecompani
64、esdevelopinggenerativeAI.In April 2023,Canada completed its third review of its Directive on Automated Decision-Making,which requires most federal institutions that use AI to carry out and publish an AlgorithmicImpactAssessment.Amendments to the directive include changing its scope to also apply to
65、internalgovernment services rather than only external ones,additional requirements for impactassessments,and a mandatory gender-based analysis of AI systems before implementing them.Canadas parliament has also made progress on Bill C-27,whichwouldamendexistingdataprivacylegislation and include a new
66、 law,the Artificial Intelligence and Data Act or AIDA.According to acompanion document released in March 2023,the AIDA would take a risk-based approach to AIregulation similar to the European Unions AI Act.The AIDA sets out high-impact AI systemswhichwould be subject to stricter requirements and wou
67、ld create an AI and Data Commissionerresponsibleforeducationandenforcement.13LookingaheadWhile the USs recent executive order represents a clear move towards regulation of AI in the US,itwill take timetoseehowtheresultingregulationsareshapedandimplemented.Inthenextyear,wecan expect a ra?of new regul
68、ations from nearly every federal department and agency as wellas NIST.Somepartsoftheexecutiveorderwillbemoredifficulttoaddressthanothers.Forexample,the order directs agencies to hire more AI experts,but competing with thehighsalariesintheUStech sector will be challenging.The executive order also urg
69、es Congress to pass comprehensivedata protection and privacy legislation,which has been introduced many times but failed thus far.While a number of states have now passed their own data privacy legislation,the lack of federallegislation covering populations in every state is a growing concern as inc
70、reasingly advanced AIsystemsrelyonlargerandlargerdatasetsthatmayputAmericansprivacyatrisk.Similarly,Canadas AIDA,if passed,would go through several rounds of dra?regulation and notcome into force until 2025.AIDAscompaniondocumentalsoexplicitlystatesthatregulationswillbedra?ed with international inte
71、roperability in mind,meaning that any regulation the US,Canadaslargesttradingpartner,adoptscouldhaveanimpactonCanadianregulatorydecisions.14GovernmentAIReadinessIndex2023Latin America&The CaribbeanBy Gonzalo Grau and Pablo Fuentes NettelTheregionallandscapeThis year,LatinAmerica&theCaribbean(LAC)pre
72、sentsanaveragescoreof41.50,placingit6thamong the 9regionscovered.Thereisasubstantialdifference,nearly40points,whencomparedtothe leading region,North America.There are also significant disparities within the region:at 63.70,thehighest-rankingBrazilisalmost42pointsaheadofHaiti,thelowest-rankingat21.97
73、.Five countries Brazil,Chile,Uruguay,Colombia,and Argentina stand out from the rest,with scores between 63.70 and 57.72 points.The performance of these countries is particularlystrong in the Government pillar,with four out of the fiveregionalleaderswithintheglobaltop40inthis category.Latin America&t
74、he Caribbean seems to be lagging in the Innovation Capacitydimension of the Technology Sector pillar,where we find a gap of almost 10 points between theregional and global average.In the Technology Sector pillar,Brazil(45.08)emerges as theregional leader,maintaining a lead of approximately 5pointsov
75、erChile(40.90)andMexico(39.55),whicharethesecondandthird-rankedcountries,respectively.15KeydevelopmentsThis year,one of the most noteworthy developments in the region was the signing of theSantiago Declaration by 20 Latin American and Caribbean governments.This document wasthemain outcome of the Sum
76、mit of Ministers and High-Level Authorities of Latin America and theCaribbean on Artificial Intelligence,organised by CAF(Development Bank of Latin America),UNESCO,and the Chilean government.The declaration acknowledges the need for proactiveinvolvementfromgovernmentstoharnesstheopportunitiesthatAIb
77、ringswhileaddressingitsrisks.This is a relevant step in terms of regional collaboration on AI readiness.The increased awarenessfostered by these initiatives could potentially lead to more robust governance and regulatoryframeworksintheregion.Another significant development within the region was the
78、Dominican Republic emerging as thefirst Caribbean country to release its National AI Strategy.This strategic document outlines aseries of initiatives aimed at integrating AI to improve public services in key sectors such asjustice,health,education,security,and transportation.Cuba and Jamaica have al
79、so announcedforthcoming AI strategies.This rise in momentum among Caribbean countries is particularlyencouraging,especially given they currently trail continental peers by 10 points in theGovernmentpillar.In a noteworthy approach to AI governance,the Mexican legislative branch is making proactiveeff
80、orts to lay the groundwork for a comprehensive AI governance framework.This initiativetakes form in the National Alliance for AI,a programme presented in the Senate and focused onstrengthening the AI ecosystem within the country.In addition,the Mexican Congress has recentlyapproved a reform to its L
81、aw on Science and Technology,paving the way towards the adoption ofethicalprinciplesfortheuseofAI.We have also seen relevant developments related to the Technology Sector pillar,the regionslowest-scoring pillar.The Dominican Republic announced innovation and capacity-building intechnology as nationa
82、l priorities in its 2030 National Innovation Policy.Similarly,Argentina islooking to spur technological innovation byopeningtwonewfinancingstreamsforAIdevelopmentandapplicationprojectsaninvestmenttotallingUSD13.5million.LookingaheadIn the upcoming year,it will be crucial to follow Brazils legislativ
83、e developments closely Brazilsproposed AI bill could be a groundbreaking initiative,positioning the country as a pioneer in AIregulation within the region.Similarly,it will be interesting to follow the latest developments ofthe EU-LACDigitalAlliance.ThecollaborationbetweentheEuropeanUnionandLatinAme
84、rica&TheCaribbean in converging digital policy and regulatory frameworks on AI,holds the potential tosignificantlyinfluencetheregulatorylandscapeintheregion.16GovernmentAIReadinessIndex2023Additionally,relevant efforts are underway in Chile and Uruguay,as they actively revise theirnational AI strate
85、gies,with support from UNESCO.These collaborative efforts could lead tostrengthening governance frameworks and aligning national strategies with international ethicalstandards.17Western EuropeBy Livia MartinescuTheregionallandscapeWestern Europe holds the second-highest global average score,with its
86、 countries comprising morethan half of the top 20 nations in this years index.It ranks behind only North America,a regionwith just two members that both rank in the top five of the Government and Technology Sectorpillars.The regionsaveragescoreof66.72reflectsaconsistentlyhighlevelofperformanceacross
87、allpillars.Similar to last years index,the United Kingdom,Finland,andFrancecontinuetoleadscoreswithin the region.Excluding the microstates Liechtenstein,Monaco,San Marino,and Andorra Western Europe exhibits a relatively narrowrangeofscorescomparedtootherregions.Greece,the lowest-ranked of the non-mi
88、crostates,is only 20.62 points behind the United Kingdom,theregional leader.Only North America,which only contains two high-ranking countries,has asmallerrangethanWesternEurope,andmostotherregionshaverangesatleast10pointslarger.As observed in last years edition of the index,this similarity in scores
89、 couldbeinpartattributedtoregional collaboration in advancing AI readiness,with regulatory and funding initiativesconcentratedattheEuropeanUnion(EU)level.KeydevelopmentsSome of the most noteworthy events in AI readiness within Western Europe over the past yearrevolve around the ongoinglegislativedev
90、elopmentsconcerningtheEuropeanUnionsAIAct.Uponfull approval,this legislation could mark Europes first dedicated AI law.The AI Act introduces18GovernmentAIReadinessIndex2023regulatory innovation by proposing a risk-based approach and imposing legal obligationsproportionate to the level of risk associ
91、ated with AI systems.By delineating limits andconditions,the legislation will play a pivotal role in shaping the aspirations of EU member statesconcerningtheintegrationofAIwithintheirnationalgovernments.In addition to these legislative strides,a noteworthy event contributing to the harmonisation ofp
92、olicy and regulation was the AI Safety Summit hosted by theUK.Thissummitledtothecreationof the Bletchley Declaration,in which 28 countries(8 inWesternEurope)andtheEUrecognizedthevariousrisksassociatedwithAIandagreedtoconveneforfurtherdiscussionsin2024.The region has also seen relevant efforts to enh
93、ance AI governance at the national level.Spainhasmade significant progress towards a more robust governance and regulatory framework.Thecountry has launched an AI sandbox to assist start-ups in adjusting to regulation whileapprovingalaw to establishthefirstsupervisoryagencyforartificialintelligencei
94、ntheEUtheSpanishAgencyfor the Supervision of Artificial Intelligence(AESIA).The AESIA will monitor adherence to AIstandards by implementing quality and responsibility seals.Additionally,it will be tasked withcreating controlled testing environments that facilitate the responsible introduction of inn
95、ovative,high-risk,or general-purpose AI systems.In a similar vein,the Netherlands has appointed anAlgorithmSupervisionBodytostrengthenthehandlingofalgorithmsthatprocesspersonaldata.Several countries in the region havecreatednewinitiativestosupportAIstart-ups,potentiallyinfluencing future scores in t
96、he Technology Sector Pillar.In 2023,the United Kingdomestablished the Frontier AI Taskforce,an internal start-up within the UK government.Theoverarching objective is to secure funding for a novel government-industry taskforce,ensuringsovereign capabilities and widespread adoption of safe and dependa
97、ble foundational AI models.This strategic move aims to solidify the UKs position as a science and technology superpower bythe year 2030.In September 2023,the Frontier AI Taskforce released its inaugural progress report,outlining key milestones.Meanwhile,Italy is launching a Corporate Venture Capital
98、 Fund for AIstart-ups to developsolutionsforthepublicsectorwiththeaimofimplementingtheir2021NationalAI Strategy and further developing AI-driven policies and services in the public sector by boostingpublicsectorinnovation.There has also been activity in creating AI centres and hubs in the region.Ita
99、ly recentlyestablished a Future Artificial Intelligence Research(FAIR)Centre to implement its policiessupporting AI.MaltahasopenedaEuropeanDigitalInnovationHub,joininganetworkof227similarhubs across Europe,which will focus on AI and otheremergingtechnologies.SpecifictoAIinpublicservices,Portugal has
100、 inaugurated a Hub for Artificial Intelligence and Data Science for PublicAdministration(AI4PA).In 2023,Western European microstates have alsorecordedimportantprogressindatacoverage.San Marino and Liechtenstein have surpassed the 50%threshold for data coverage this year,anotable improvement from the
101、 previous year where their data representation was less extensive.33Countries with data coverage below 50%of indicators are excludedfromtherankings.Thisyearsincreasein data availability has reduced the number of countries omitted from the ranking to only two this year,asignificantimprovementfromlast
102、year.19While the current data coverage for both countries stands at 53.85%,meaning their scores areheavily reliant on imputed data,itrepresentsapositivetrend.Despitethesuboptimalcoverage,thisdevelopment signifies a step in the right direction,showcasing an encouraging trend towardsenhanceddatainclus
103、ivity.LookingaheadEfforts are underway to narrowthedisparitybetweentheprogressobservedintheGovernmentandData and Infrastructure pillars and the relatively lower scores intheTechnologySectorpillar,whereWestern European countries exhibit their weakest performance.A case in point is Germany,whichhas de
104、monstrated notableadvancementsthroughitsAIActionPlanforeducationandresearch.Thisstrategic initiative is designed to elevate AI infrastructure,with a specific emphasis on computerinfrastructure enhancement.Concurrently,the plan aims to foster skill development in the field ofAI.Furthermore,it will be
105、 important to follow closely the legislative developmentssurroundingtheAIAct,asitsapprovalcouldhaveamajorimpactonhowAIisgovernedbyEUmembers.Furthermore,it will be crucial to closely monitor the latest developments related to theapprovaloftheEUAIAct,particularlyregardingtheregulationoffoundationmodel
106、s.Foundationmodels,such as those empowering large language models like ChatGPT or Google Bard,areunderscrutiny.The amendment proposed by France,Germany,and Italy advocates for codes of conduct andself-assessment for foundation models without an initial sanction regime,contrasting withprescriptive ob
107、ligations in the original dra?.This modification may impact how governmentscollaboratewithtechcompaniestointegrateGenAItoolsforenhancingpublicservices.20GovernmentAIReadinessIndex2023Eastern EuropeBy Livia Martinescu and Emma HankinsTheregionallandscapeThe average score across the Eastern European c
108、ountries this year is 54.67,meaning the regionranks third globally a?er North America and Western Europe.Eastern Europe,on average,lagsbehind Western Europe by 12.05 points,suggesting a clear gapacrosstheboardbetweenthesetworegions.Leading the charge is Estonia,securing the top position with a score
109、 of 70.86 and ranking17th globally.Czechia ranks second with a score of 65.17 and Lithuania holds the third spot with ascoreof63.33,withPolandandtheRussianFederationclosebehindat63.10and62.92,respectively.There is a large range of scores within this region regional leaderEstoniais34.38pointsaheadofl
110、owest-ranking regional player Bosnia and Herzegovina demonstrating clear disparities in AIreadiness in Eastern Europe.Estonia also stands out from the other top-ranking countries in theregion,with the 5-point gap between it and second-ranking Czechia larger than any other gapbetweentwonext-rankingco
111、untriesintheregion.KeydevelopmentsEstonia,which is world-leading in many e-government indicators,is in the process of launching itsmRiik digital identity app,which wasdevelopedwiththeUkrainiangovernmentandmodelleda?erits Diia app.Estonia is also actively contributing to global standards for the digi
112、tisation of publicsector infrastructure through its support of GovStack,an open-source community that provides21technology,technical specifications,and implementation support for government digital services.This way,Estonia is using policy innovations to support the integration of AI into the govern
113、ancesystems in the Eastern European region andbeyond.In2023,GovStackreleasedanupdatedversionof the Building Blocks Specifications,accompanied by the GovStack Implementation Playbook.These resources offer valuable guidance for governments and technology developers,emphasisingcost-effectiveandvendor-a
114、gnosticsolutionsfordeliveringe-governmentservices.The Eastern European region has seen several promising initiatives in the AI regulationlandscape.The Ministry of Digital Transformation in UkraineunveiledaroadmapfortheregulationofAI.ThisroadmapisdesignedtoassistUkrainiancompaniesinreadinessfortheimp
115、lementationoflegislation akin to the EuropeanUnionsAIAct.Meanwhile,theSerbiangovernmentadoptedEthicalguidelines for the development,application,and use of reliable and responsible artificialintelligence,which it recommendsallpublicbodiesfollow.Theseguidelinesarenotablefornotonlylisting broad ethical
116、 principles for responsible AI but also including detailed questions for AIdevelopers and public servants to answer,acting as a self-assessment tool for whether AI systemsare being developed and used responsibly.This makes SerbiaonlythesecondcountryintheregiontohavepublishedethicalAIguidelines,a?ert
117、heRussianFederation.Some of the European Union members in the region(Bulgaria,Croatia,Czechia,Estonia,Hungary,Latvia,Lithuania,Poland,Romania,Slovakia,and Slovenia)have benefited frominvestments supported by the EUs Recovery and Resilience Facility,which funds projects andreforms to prepare member s
118、tates for the green and digital transitions.For example,Latviaisusingthese funds to establish the Latvian Federal Cloud,while Romania has used them to modernisepublic libraries and turn them into digital skills hubs.The Lithuanian government has combinedthese funds with a number of other investments
119、 in AI and digital transformation,including:afunding opportunityof15millionforAI,blockchain,androboticsstart-upsandspin-offssupportedby the Recovery and Resilience Facility;6 million for development of digital services using opendata;andfundingworth115.26millionfordigitalisationofpublicsectorservice
120、s.LookingaheadEU support and investments like the ones above might provide one explanation for thedisparityinscores between the highest-and lowest-scoring countries in the Eastern European region:the fivelowest-ranking countries are all non-EU members,whileallofthetopfivecountriesexcepttheRussian Fe
121、deration are EU members.However,the lower-income,non-EU states of Ukraine andSerbia manage to outrank the high-income EU member states Romania and Croatia.This is largelydue to Ukraine and Serbias higher scores in the Government pillar;Romania and Croatia are theonly EU member countries in Eastern E
122、urope that still lack national AI strategies.These will becountries to watch as the EU finalises its AI Act.Will EU member states rely on EU-wide regulationinstead of creating their own AI strategies?And how will the AI Act affect non-member states likeUkraineandSerbia?There are also open questions
123、about whether the region can catch up in its lowest-scoringpillar,the Technology Sector.The regions average scores in each dimension of thispillarareonlyslightly above the global average,and are 10 to 16 points behind WesternEuropeanaverages.Even22GovernmentAIReadinessIndex2023the three regional lea
124、ders lag behind their counterparts in other regions,withtop-rankingEstoniasTechnology Sector pillar score below that of the top countries in North America,Western Europe,and East Asia.Unless Eastern European countries catch up with those in other regions,they facevulnerability across various sectors
125、,impacting growth,competitiveness,sustainability,inclusion,security,andstrategicstrength.23Middle East&North AfricaBy Livia MartinescuTheregionallandscapeThe Middle East and North Africa(MENA)region displays a significant variation in government AIreadiness,ranking as the region with the third-large
126、st range of scores.There is a clear disparitybetween Middle Eastern and North African countries,with average scores of 38.89 and 51.11,respectively.Notably,however,Egypt is a North Africanoutlier,securingapositionamongthetopten in MENA,primarily attributed to itsgoodperformanceintheGovernmentpillar.
127、TheUnitedArabEmirates(UAE)takes the lead in the region,driven by relatively high scores across all three pillarsthatplaceit18thworldwide.KeydevelopmentsIn 2023,the MENA region witnessed significant developments in terms of governance and AIethics principles.Egypt has made notable advancements in thi
128、s domain by introducing theEgyptian Charter for Responsible AI.This initiative combines insights with actionable measures tofacilitate the responsible development,deployment,management,and utilisation of AI systems.Adapting guidelines established by internationalorganisations(OECD,UNESCO,WHO,IEEE,EU
129、),theCharter aims to foster awareness among all stakeholders in the AI ecosystem regarding ethicalconsiderationsinAI.24GovernmentAIReadinessIndex2023Similarly,the Kingdom of Saudi Arabia(KSA)published its AI Ethics Principles.This framework isaimed at reducing the potentially negative implications o
130、fAIsystemsandprotectingandenhancingprivacy and personal data protection.The AI Ethics Framework of the KSA outlines principles forgoverning AI and includes a risk classification system associated with the implementation of AIsystems,similar to the European Unions proposed AI Act.Through the applicat
131、ion of theseframeworks,the government aims to ensure that AIdecision-makingprocessesupholdfairnessandequityforitscitizens.Recently,countries in the MENA region have taken important stepstowardsimprovingtheirData&Infrastructure foundations.Bahrain has released its Sixth National Telecommunications Pl
132、an,outlining the governments strategic vision and overall policy for the telecommunications sector.Within this document,Bahrain emphasises the critical need for establishing resilient infrastructureand connectivity foundations to effectively integrate technologies like AI.The initiative recognisesth
133、e imminent impact of rapid technological change on both telecommunications and the broaderICT market structure.Furthermore,this document underscores Bahrains plans to formulate anationalAIstrategy.The region has also undergone a notable surge in data centre investments.Recently,Huaweiannounced the i
134、nauguration of Riyadhs cloud region,and Oman established a partnership withSAP to introduce a private cloud data centre.Furthermore,Egypt is set to host a hyperscale datacentre with a substantial investment of$250 million.These advancements are poised to influencethe regions readiness in terms of Da
135、ta&Infrastructure,a pillar where they currently score 4pointsbelowtheglobalaverage(60.09).LookingaheadThe effective integration of AI into government operations depends on having a skilled workforce.Looking forward,a potential area for advancement in the MENAregionisHumanCapital,wheretheaverage scor
136、e stands at 44.70.In this context,the region showcases promising initiatives thathave the potential to cultivate a more proficient workforce.Notably,the United Arab Emirateshas launched the Coders(hq)programme,designed to establish connections between global techinfluencers and the local tech commun
137、ity through events such as Hackathons,conferences,andtrainingprograms.Moreover,it is pertinent to closely observe the AI strategy initiatives in the region,with threecountries Iraq,Tunisia,and Bahrain announcing upcoming vision documents.This marks acrucial step in enhancing the positioning of these
138、 countries within the landscape of government AIreadiness.The successful implementation of these initiatives would lead totheMiddleEast&NorthAfrica joining Western Europe,Eastern Europe,East Asia,and North America as one of the regionswheremorethanhalfofthecountrieshaveintroducedAIstrategies.25Sub-S
139、aharan AfricaBy Emma HankinsTheregionallandscapeDespite Sub-Saharan Africa having the lowest average score of any world region in the index,signifying serious challenges to government AI adoption in the region,there hasbeenrealgrowthover the past 12 months,with 3 countries publishing new national AI
140、 strategies and oneannouncing a forthcoming strategy.In addition,3 countries have announced they areworkingwithUNESCOtoadoptandimplementstrategiesinlinewithUNESCOsRecommendationontheEthicsofAI.Mauritius leads the region with a score of 53.27,followed by South Africa,Rwanda,Senegal,and Benin in the t
141、op five.Mauritiuss strength lies mostly in the Government pillar its score of69.82 in this pillar is 10 pointshigherthanitsscoreinanyotherpillar whileSouthAfricaleadstheregion inboththeData&InfrastructurepillarandtheTechnologySectorpillar.Infact,SouthAfricaisthe only country in Sub-Saharan Africa to
142、 score above the global averagefortheTechnologySectorpillar.26GovernmentAIReadinessIndex2023KeydevelopmentsDespite these barriers,the past year has seen significant and interesting developments ingovernment AI readiness in the region.In particular,Rwanda,Senegal,and Benins new nationalAI strategies
143、representthefirstAIstrategiesinmainlandSub-SaharanAfricaandendMauritiuss5-year term as the only country in the region with an AI strategy.It is also worth noting that thesecountries are all considered low income(Rwanda)or lower middle income(Senegal and Benin)bythe World Bank,unlike upper middle inc
144、ome Mauritius.This goesagainstthetrendwehaveseeninsome other world regions,in which the most developed or largest economies are the firsttocreatenational AI strategies.By this logic,countries like South Africa or Nigeria would create national AIstrategiesfirst,yetneitherhasdoneso(thoughNigeriasstrat
145、egyisreportedlyindevelopment).It is also worth noting that Rwanda and Senegal both createdtheirAIstrategieswithsupportfromcooperation agencies and international organisations GIZ FAIR Forward,the World EconomicForum,and The Future Society in Rwandas case;and the African Union and Team Europe via the
146、AU-EU Digital for Development(D4D)Hub in Senegals case.Furthermore,regional collaborationseemstobeagrowingtrendintheregion,withNamibiahostingministersfromBotswana,Malawi,Mozambique,SouthAfrica,Zambia,and Zimbabwe at the first UNESCO-Southern Africasub-Regional Forum on Artificial Intelligence(SARFAI
147、).This forum produced the WindhoekStatement,which recommended actions on AI governance,capacity-building,infrastructure,R&D,environmental protection,gender inclusion,and collaboration across the region.Internationalcollaboration is also visible in Cte DIvoire,Namibia,and Rwanda,which have(separately
148、)committed to working with UNESCO on implementing the Recommendation on the Ethics of AI.Meanwhile,GIZ FAIR-Forward and the Global Partnership for Sustainable Development Datacollaborated with Kenyan stakeholders to co-create an AI Practitioners Guide specific to Kenyaslegalandregulatoryenvironment.
149、While most countries in the region lack AI strategies,there has been continued progress in dataprotection policies and government digital transformation,which are essential foundations forgovernment AI readiness.Nigeria enacted anewDataProtectionActin2023,replacingits2019DataProtection Regulation wi
150、th more comprehensive legislation,and Senegal published its NationalData Strategy,developed with the nonprofit Smart Africa and German development agency GIZ.International cooperation in digital transformation is likely to continue,as the US announced aDigital Transformation with Africa initiative,t
151、he UK launched an AI for Development programmefocused on Sub-Saharan Africa,and GIZ operates eleven Digital Transformation Centres in theregionandhaspossibleplanstoexpand.LookingaheadIt appears the increase in national AI strategies in Sub-Saharan Africa will continue next year,withEthiopia and Nige
152、ria announcing work on dra?AI strategies.Perhaps the most anticipated future27development is the publication and approval of the African Unions AI Continental Strategy forAfrica,expected to be launched at the January 2024 AU Summit.As Rwanda and Senegals newstrategies illustrate,international organi
153、sations can be influential in supportingAfricannationsto cra?their own national AI strategies.A continental strategy from the African Union couldprovide valuable guidance and a blueprint for other countries to follow while savingthemprecioustimeandresourcesindra?ingandimplementinganAIstrategy.28Gove
154、rnmentAIReadinessIndex2023South&Central AsiaBy Livia MartinescuTheregionallandscapeThe South and Central Asia region registers the second-lowest average globally,with a score of40.17.The region also exhibits a considerable range of 44.46 fourth-highest in the world andsimilar to that of the Middle E
155、ast and North Africa and Pacific regions.The diverse scores acrosscountries in the region indicate significant disparitiesintheirreadinessforAItransformation.Thisispartially attributed to major differences in economic development,technological adoption,andgovernancestructuresamongthecountries.India
156、and Trkiye stand out as the top performers in the region,securing the first and secondpositions respectively,and leaving an 11-point gap between these countries and third-rankingKazakhstan.On a global level,India and Trkiye demonstrate competitive scores in both theGovernment pillar,as both rank int
157、hetop30countriesworldwide,andtheTechnologySectorpillar,where bothrankinthetop50globally.However,bothfallbehindintheDataandInfrastructurepillarglobally,withTrkiyeranking61standIndia66th.KeydevelopmentsThe regional leader,India,released the first edition of IndiaAI 2023,dra?ed by an Expert Group forth
158、e Ministry of Electronics and Information Technology.The detailed report serves as an29implementation plan for Indias 2018 national AI strategy and provides a roadmap for theintegration of AI into the structure of the nation,encompassing governance,data management,and strategic partnerships to foste
159、r innovation and technological advancement,and serves as acornerstone for the nations AI readiness.The publication offers strategic recommendations tocapitalise on Indias demographic dividend thatis,thepotentialforeconomicgrowththatoccurswhen a labourforcegrowsfasterthanthepopulationdependentuponita
160、ndleverageitsstatusasan IT superpower.Alongside this,India is providing ever more use cases of AI in the public sector,including launching Bhashini,an interactive platform that aims to enable citizens to access theinternet and digital services in their own language as part of the Indian governments
161、NationalLanguageTranslationMission.This year,Trkiye published its Industry and Technology Strategy,articulating its aspirations toemerge as a technological power and emphasising a synthesis of competitiveness,autonomy,andinnovation.Amidst these ambitions,Trkiye also aims to further its domestic digi
162、tal policies byaddressing data governance,content moderation,and competition policy.Moreover,the strategyprovides recommendations related to the use of AI,particularly in relation to dataprotectioninthefield.The strategic alignment of these policies signals Trkiyes aim of fostering an environmentcon
163、ducivetotechnologicaladvancementandinnovation.Central Asian countries have also shown progress in AI readiness and digital transformation.Tajikistan has published its national AI strategy,which seeks tobothencourageAIcompaniestosetup in Tajikistan and adapt the countrys legislative and regulatory fr
164、ameworks to support its goalsfor AI.Meanwhile,in Kazakhstan,the Astana Hub,an international technologyparkwithspecialtaxand visa status,has partnered with Google for Startups on its Silkway Accelerator programme toprovidesupportandmentoringfortechstartupsintheregion.While it may seem these countries
165、canbeovershadowedbythemuchlargereconomiesofIndiaandTrkiye as well as powerful neighbouring countries like the Russian Federation and China,CentralAsian countries are reportedly seeking a common,harmonised approach to AI in the region,possiblythroughtheC5groupofTajikistan,Kazakhstan,Kyrgyzstan,Turkme
166、nistan,andUzbekistan.LookingaheadOverall,the region presents low average scoresinspecificdimensionsacrossallthreepillarsofour index.Some,like the Maturity dimension of the Technology Sector pillar,may improve withtime,especially given some countries stronger scores inHumanCapitalindicatorslikeGradua
167、tesinSTEM or computer science.Other areas,however,may be cause for concern in the future.Forexample,the region has low average scores on Data Availability and Data Representativeness,twocriticaldimensionsofcreatingunbiased,relevantAItools.The year ahead may be a busy one for AI governanceinSouthandC
168、entralAsia.RegionalleaderIndia will take council chairmanship of the Global Partnership on Artificial Intelligence(GPAI),amulti-stakeholder initiative focused on AI governance and responsible AI.The region also has anumber of forthcoming strategies:Azerbaijan,Uzbekistan,andSriLankahavestatedtheirint
169、entionto develop strategies,and Pakistan has released a dra?document.These and a possible C530GovernmentAIReadinessIndex2023agreement on AI could be influential in a region in which 78%of countries still lack national AIstrategies.31East AsiaBy Gonzalo Grau and Emma HankinsTheregionallandscapeEast A
170、sia presents an average score of 51.41,placing it 4th amongst the 9 regionscovered.Lookingclosely,the region presents a higher score than its immediate neighbours.Itscores11pointsaboveSouth and Central Asia and outperforms the Pacific by 10 points.The regions high-scoringperformance,however,hideslar
171、geregionaldisparities:EastAsiahasthehighestrangeofscoresingovernment AI readiness this year,with 72 points between first-ranking Singapore and18th-ranking Democratic Peoples Republic of Korea(DPRK).Even excluding the DPRKs low score,which is partially due to lack of data,there is a 52-point gap betw
172、een Singapore and 17th-rankingTimor-Lestestillthelargestrangeinscoresofanyregion.In a region where more than one third of countries are below the global average of 44.94,regionalleaders like Singapore(81.97),the Republic of Korea(75.65),and Japan(75.08)are performingexceptionally across all pillars.
173、All three economies score above 80 points out of 100 in both theGovernment and Data and Infrastructure pillars.With regards to the Technology Sectorpillar,thesecountries along with China,Taiwan,and Malaysiaareamongthetop25performersworldwide.Most notably,Singapore is the global leader in both the Go
174、vernment(90.40)and the Data andInfrastructure(89.32)pillars.32GovernmentAIReadinessIndex2023KeydevelopmentsThere has beensignificantcollaborationbetweenregionalleadersandcountriesoutsideofEastAsia in the past year.Most prominently,Japan has held the G7presidencyfor2023andconvenedasummit in Hiroshima
175、 that started a processofdevelopingresponsibleAIguidelines.TheRepublicofKorea agreed a memorandum of understanding with the United Kingdom on strengtheninggovernment digital services,including exploring the potential of AI in government.Thismemorandum is part of a larger science and technology accor
176、d between the two countries,whichincludes a commitment to coordinating the development of responsible AI as well as a frameworkforcooperationonsemiconductors.East Asia has also seen one of the worlds first laws on generative AI come into effect.InAugust,the Cyberspace Administration of China(CAC),al
177、ong with six othergovernmentagencies,approvedthe Interim Measures for the Administration of Generative Artificial Intelligence Services,whichrequire firms to follow rules on data privacy and intellectual property,but only for generative AIsystems that wouldbeavailabletothepublic.Thesemeasurescomeast
178、emporarylegislationbeforean all-encompassing national AI Law is dra?ed.Elsewhere,Singapores Monetary Authority hasbegun collaboration with banks and tech firms to develop a riskframeworkforgenerativeAI,whileits government has begun trialling a ChatGPT-like chatbot for public servants and testing how
179、similarlargelanguagemodels(LLMs)canimproveusersexperiencesaccessingsupportservices.Large tech companies are also takinganinterestintheregion,providingfundsforupskillinginmultiple countries.Mongolia has signed a digital skills partnership with Google in order toaccelerate its human capital developmen
180、t,while Singapore is working with Microso?totrainSMEsin data use and protection.Malaysias government also announced a partnership with Google totrain individuals in AI and data analytics for free and the company is considering the creation of aGoogledatacentreinMalaysia.There have also been promisin
181、g developments in East Asian countries beyond the regionalleaders.In its first government report on AI in the country,Cambodian scholars made detailedrecommendations on how the government could support and regulate AI,potentially laying thegroundwork for a strategy.Amidst a series of moves to modern
182、ise its regulatory framework forincreasing AI deployment and uptake,Papua New Guinea has published the third iteration of itsData Governance and Protection Policy.It hasalsobeguntoreformitsnow14-year-oldNationalICTAct in an effort to adapt to the current technological landscape,a move which has been
183、accompaniedbyalargerfocusonSMEsupportwithinitsnationalbudgetfor2023.LookingaheadChinas in-progress AI law adds to a number of potentially influential AI governance initiativescurrently in the dra?stagesinEastAsia.OnemajorinitiativeistheAssociationofSoutheastAsian33Nations(ASEAN)s so-called ASEAN Gui
184、de on AI Governance and Ethics,which tech companies arereportedly being consulted on and may be finalised as soon as January 2024.While no dra?ofthisguide,which would be voluntary,has been made public,reports suggest that it may take a lessprescriptive approach than the EUs AI Act.On the other hand,
185、Japans dra?AI guidelines willreportedly require organisations that use the technology to ensure they are not relying too heavilyon AI and maintain records of their interactions with AI.Whatever the final versions of thesedocuments contain,the policies of a large organisation like ASEAN andaregionala
186、ndgloballeaderin government AI readiness like Japan could certainly influence the direction of future national AIpolicyintheregion.34GovernmentAIReadinessIndex2023PacificBy Livia Martinescu and Emma HankinsTheregionallandscapeThis year,the Pacific region attains the seventh-lowest scores,with an ave
187、rage of41.33andarangeof 46.48.Our data shows an evident regional gap in AI readiness.There is a noticeable disparitybetween the leading economies in the region(Australia and New Zealand)and the rest of thePacific nations,which are Small Island Developing States(SIDS):we see a difference of 13.44 ins
188、core between the second and third highest-ranking nations in the region.The Pacific SIDS,characterised by geographical dispersion and relatively small populations,face the challenge ofbeingdistantfrommajorinternationalmarkets.Globally,Australia secures the 12th position,whereas New Zealand holds the
189、 49th spot.Notably,both countries excel in the Data and Infrastructure pillar,achieving scores of 85.75 and 81.66respectively.These impressive scores position Australia at 3rd place and NewZealandat13thplaceonaglobalscalewithinthispillar.This year,the number of countries included intheindexwithinthe
190、Pacificregionhasincreasedfrom eight to eleven.This expansion signifies progress in data availability,asourindexexclusivelyevaluates countries with values for more than 50%of our indicators.Despite this positivedevelopment,a considerable portion of the scores for these additional countries is still d
191、ependent35on imputed data,given their proximity to the inclusion cutoff.Further efforts are imperative toenhancetheavailabilityofdata,particularlypertainingtogovernmentandeconomicaspects,acrosstheregion.KeydevelopmentsRegional leaders Australia and New Zealand havebothseendevelopmentsinAIgovernancet
192、hisyear.In the Australian Governments response to the Australian Parliaments Inquiry into SocialMedia and Online Safety,it announced that it will undertake research into algorithms to improveunderstanding of the use and harms related to algorithms on digital platforms and determine ifthere are regul
193、atory gaps for the Governments consideration.This initiative will drive regulatorymeasures and reforms for the Australian government in the application of AI technologies.Inaddition,eSafety,Australias independent regulator and educator for online safety,published aposition statement addressing gener
194、ative AI.The statement outlines various regulatory challengesandproposedapproachestoaddressingthisevolvingtechnology.Likewise,second-ranking New Zealand published interim generative AI guidance for the publicservice in July 2023,and the Ministry of Education published guidelines for teachers on usin
195、ggenerative AI in schools.Notably,these guidelines are the first AI policy documents NewZealands government has released publicly,and they recommend that each agency develop itsown AIpolicy,whileprovidingsomerecommendationsonmitigatingtherisksofgenerativeAI.Whilenew government AI policies are a welc
196、ome change for New Zealand,which still lacks a national AIstrategy and ethical AI guidelines,public servants may need to work across agenciestoensurethatgovernment AI policies are consistent and clear to service users.The government has recognisedthis,directing the Ministry for Business,Innovation a
197、nd Employment to lead development of across-agencyworkprogrammeonAIandreportbacktoministersbyearly2024.New Zealands government also recently established the Interim Centre for Data Ethics andInnovation to support safe,inclusive,and innovative applications of data.The centre prioritisesleading the re
198、sponsible use of data and promoting ethical,data-driven innovation.Its objectivesinclude cultivating public trust and confidence while enabling government agencies to engage ininnovativepracticeswithdata.Beyond Australia and New Zealand,the Pacific island nations AI and digital transformationefforts
199、 largely focus on addressing and mitigating the existential threat of climatechange.Forinstance,Tuvalus government has created the Future Now Project,which,among other things,seeks to digitise government administration and important historical and cultural documents toprotect them in the face of mor
200、e frequent natural disasters.Meanwhile,the United Nations CapitalDevelopment Fund is assisting Fiji in harnessing the power of AI in natural disaster recovery.Notonly is this programme accelerating the restoration of local livelihoods,it also increases access todigitaltoolsinanareathatisaffectedbydi
201、gitaldivides.36GovernmentAIReadinessIndex2023LookingaheadOver the past few years,SIDS have made significant strides in enhancing their readiness for AI.Global leaders will gather in Antigua&Barbuda in 2024 toestablishanewactionplanoutliningthedevelopment priorities for SIDS,including the Pacific isl
202、and nations,over the next decade.Thispresents Pacific island nations with a unique opportunitytodelineatetheirprioritiesintherealmofAI.Simultaneously,it allows them to present a united and influential front among low and middleincome countries(LMICs),taking a prominent role in international AI gover
203、nance discussions toensure that AI is governed inclusively.Both Australia and New Zealand should also see progressonAI governance in the next year,with Australias research into the risks of algorithms and NewZealandscross-agencyworkprogrammebothduetobedeliveredinearly2024.37Annex I:MethodologyDimens
204、ionsandIndicatorsThetablebelowsummarisesthedimensionsandindicatorsusedintheIndex.Government PillarDimensionDescriptionIndicatorSourceVisionDoesthegovernmenthaveavisionforimplementingAI?NationalAIstrategy(Y/N)Deskresearch(e.g.OECDAIPolicyObservatory,UNIDIRAIpolicyportal)GovernanceandEthicsAretherethe
205、rightregulationsandethicalframeworksinplacetoimplementAIinawaythatbuildstrustandlegitimacy?DataprotectionandprivacylegislationUNdataprotectionandprivacylegislationworldwideCybersecurityGlobalCybersecurityIndexRegulatoryqualityWorldwideGovernanceIndicatorsNationalethicsframework(Y/N)Deskresearch(e.g.
206、Nature,AIEthicsLab)AccountabilityWorldwideGovernanceIndicatorsDigitalCapacityWhatistheexistingdigitalcapacitywithingovernment?OnlineservicesUNe-GovernmentSurveyFoundationalITinfrastructureWorldBankGovTechMaturityIndexGovernmentpromotionofinvestmentinemergingtechnologiesNetworkReadinessIndex38Governm
207、entAIReadinessIndex2023AdaptabilityCanthegovernmentchangeandinnovateeffectively?GovernmenteffectivenessWorldwideGovernanceIndicatorsGovernmentsresponsivenesstochangeGlobalCompetitivenessIndexProcurementdataGlobalDataBarometerTechnology Sector PillarDimensionDescriptionIndicatorSourceMaturityDoesthec
208、ountryhaveatechnologysectorcapableofsupplyinggovernmentswithAItechnologies?NumberofAIunicornsCBInsightsNumberofnon-AItechnologyunicornsCBInsightsValueoftradeinICTservices(percapita)UNCTADValueoftradeinICTgoods(percapita)UNCTADComputerso?warespendingGlobalInnovationIndexInnovationCapacityDoesthetechn
209、ologysectorhavetherightconditionstosupportinnovation?TimespentdealingwithgovernmentregulationsWorldBankWorldDevelopmentIndicatorsVCavailabilityGlobalInnovationIndexR&DspendingUNESCOCompanyinvestmentinemergingtechnologyNetworkReadinessIndexResearchpaperspublishedinAIScimago39HumanCapitalAretheretheri
210、ghtskillsinthepopulationtosupportthetechnologysector?GraduatesinSTEMUNESCOGitHubusersperthousandpopulationGitHubFemaleSTEMgraduatesWorldBankQualityofengineeringandtechnologyhighereducationQSEngineering&TechnologyrankingsICTskillsITUData and Infrastructure PillarDimensionDescriptionIndicatorSourceInf
211、rastructureDoesthecountryhaveagoodtechnologicalinfrastructuretosupportAItechnologies?TelecommunicationsinfrastructureUNe-GovernmentSurveySupercomputersTop500BroadbandqualityEIUInclusiveInternetIndex5GinfrastructureOokla5GMapAdoptionofemergingtechnologiesNetworkReadinessIndexDataAvailabilityIstherego
212、odavailabilityofdatathatcouldbeusedtotrainAImodels?OpendataGlobalDataBarometerDatagovernanceWorldBankGovTechMaturityIndexMobile-cellulartelephonesubscriptionsITUHouseholdswithinternetaccessITUStatisticalcapacityWorldBank40GovernmentAIReadinessIndex2023DataRepresentative-nessIsthedataavailablelikelyt
213、oberepresentativeofthepopulationasawhole?GendergapinInternetaccessEIUInclusiveInternetIndexCostofinternet-enableddevicerelativetoGDPpercapitaGSMAMobileConnectivityIndexMissingValuesOnlycountrieswithvaluesformorethan50%ofindicatorsareincludedinthefinalindex.Forthisreason,thefollowingcountrieswerenoti
214、ncludedinthefinalrankings:FederatedStatesofMicronesiaPalauPeerGroupMeanImputationForthemajorityofindicatorswithsomedatamissing,weimputedthevalueofthepeergroupmeanforeachcountry(wherepeergroupistheirgeographicalregionplustheirWorldBankincomegroup).Forafewindicators,theonlypossiblescoresare0,50,or100.
215、Whenweimputedscoresfortheseindicators,weroundedthepeergroupmeantothenearestpossiblescore.For11countries,imputationofpeergroupmeanswasnotpossibleforsomemissingindicatorsastheywereeitherthesolecountryintheirpeergrouporinapeergroupinwhichallcountriesweremissingdataforthatindicator.Thesecountrieswere:Af
216、ghanistanCanadaDemocraticPeoplesRepublicofKoreaIraqLibyaMaldivesSeychellesStateofPalestineSyrianArabRepublicUnitedStatesofAmericaYemen41Forthesecountries,noimputationwasattemptedwherevaluesweremissing.CalculatingScoresNormalisationAllscoreswerenormalisedtobebetween0and100.Theformulafornormalisationw
217、asasfollows:ForallindicatorsexceptRegulatoryquality,Accountability,andEffectivenessofgovernment(where=-2.5),thevalueofwassetto0.waseitherthemaximumpossiblevalue(inthecaseofdatafromotherindices,e.g.theNetworkReadinessIndex),orthemaximumobservedvalue.ThetwoexceptionswereforMobile-cellulartelephonesubs
218、criptionswherewesetthemaximumvalueto130(i.e.above130subscriptionsper100mobile-cellulartelephonesaresufficientlywidespreadtowarrantascoreof100,andvaluesanyhigherwouldnotrepresentasignificantimprovement),PercentageofSTEMgraduateswhoarefemalewherewesetthemaximumvalueto50(i.e.countriesclosertogenderpari
219、tyinSTEMgraduatesscorehigher,butonceparityisachieved,ahigherpercentageoffemalegraduatesdoesnotmeanahigherscore).Fortwoindicators,TimespentdealingwithgovernmentregulationsandGendergapininternetaccess,scoreswerenormalisedusingtheformula100,sothatahigherscoreisgiventocountrieswithasmallergendergapinint
220、ernetaccessandwherelesstimeisspentdealingwithgovernmentregulations.TreatmentofSkewedIndicatorsNineindicatorswereidentifiedasskewed(either(a)absoluteskewness2.0andkurtosis3.5or(b)kurtosis10).Thesewere:NumberofAIunicornsNumberofnon-AItechnologyunicornsValueofICTgoodstradepercapitaValueofICTservicestra
221、depercapitaVCavailabilityR&Dspending42GovernmentAIReadinessIndex2023AIresearchpapersGitHubusersSupercomputersAsVCavailabilityisapretreatedscoreratherthanharddata,thiswasle?untreated.TheindicatorsNumberofAIunicorns,Numberofnon-AItechnologyunicorns,andSupercomputershavelargenumbersofcountriesscoring0.
222、Wewerethereforewillingtotolerateahigherdegreeofskewnessintheseindicators.However,lookingpurelyatthecountrieswithavalueof0intheseindicators,absoluteskewnesswasstill2.0andkurtosisstill3.5,sowestillfelttheneedtotreattheminsomeway.TheeightskewedindicatorsweretreatedusingthelogarithmictransformationFor(1
223、+).indicatorsotherthanNumberofAIunicorns,Numberofnon-AItechnologyunicorns,andSupercomputers,thisbroughtskewnessandkurtosisdowntoacceptablelevels.Fortheotherthreeindicators,skewnessandkurtosisinthesubsetofcountrieswithvalues0werebroughtdowntoacceptablelevels,evenifskewnessandkurtosiswerestillhigherin
224、theindicatoroverall.Followingthelogarithmictransformation,theindicatorswerenormalisedasabove.TotalScoreTocalculatethetotalscore,wetookthearithmeticmeanofeachdimension.Thenwetookthearithmeticmeanofeachpillar.Thefinalscoreisthearithmeticmeanofthethreepillars.Allindicators,dimensions,andpillarswereweig
225、htedequally.NewindicatorsInpreviouseditionsoftheIndex,weusedfourindicatorsfromtheWorldEconomicForumsGlobalCompetitivenessIndex.However,thisindexhasnotbeenupdatedsince2019,andthisyearwefeltitwastimetoattempttoreplaceasmanyoftheseindicatorsaspossiblewithmoreup-to-datedata.WealsodecidedthattheCloudprov
226、idersindicatorfromlastyearsIndex,whichmeasuredwhethercountrieshadaccesstocloudservicesfromjustthetop5cloudprovidersbymarketshare,wasunnecessarilyexclusivetoothercloudprovidersandthusnotthebestmeasureforcloudcapacityinagivencountry.Forthisreason,wedecidedtoreplaceitwithanindicatormeasuringthenumberof
227、top500supercomputerslocatedinagivencountry,asaproxyforcomputepoweravailableinthatcountry.43Old indicatorNew indicatorDescription of new indicatorLegalframeworksadaptabilitytodigitalbusinessmodelsRegulatoryquality(WorldwideGovernanceIndicators)Measuresperceptionsoftheabilityofthegovernmenttoformulate
228、andimplementsoundpoliciesandregulationsthatpermitandpromoteprivatesectordevelopment.BusinessadministrativerequirementsTimespentdealingwithgovernmentregulations(WorldBankEnterpriseSurveys)Percentofseniormanagementtimespentdealingwiththerequirementsofgovernmentregulations.DigitalskillsICTskills(ITU)Pe
229、rcentofthepopulationwithICTskills,definedashavingcarriedoutaspecifictaskinthepastthreemonths.Thereisdataonelevenskills,butthreewereexcludedduetolowcountrycoverage.Anaverageofthepercentagesfortheremainingeightskillswastakentocreateonescorepercountry.Ifacountrywasmissingdataonfourormoreskills,itwasnot
230、givenascore.CloudprovidersSupercomputers(Top500)NumberofsupercomputerslistedontheTop500listofsupercomputersintheworld.Wedecidedtokeeponeindicator,Governmentresponsivenesstochange,fromthe2019GlobalCompetitivenessIndexbecausewewereunabletofindasuitablealternativeindicator.Wearehopefulthattheforthcomin
231、gfirsteditionoftheWorldBanksBusinessReadyreportwillprovidemoreup-to-dateandrobustalternativestotheseindicatorsinthefuture.LimitationsoftheDataAgeofthedataWhile the majority of our datasets were from 2023,2022 or 2021,some were older.Aside from thesingle indicator we kept from the 2019 Global Competi
232、tiveness Index,the indicators Female STEM44GovernmentAIReadinessIndex2023graduates and Statistical capacity are based on data from 2019,while the Cybersecurity indicatordata is from 2020.While we recognise this means that countries scores may not reflect the mostup-to-date picture,we still chose to
233、include these datasets given their importance andthelackofasuitablealternative.The following indicators included data from a range of years and some countries had no data forrecentyears.Whereacountryhadavaluemissing,weusedthemostrecentperiodforthatvalue.ValueoftradeinICTservicespercapitaValueoftrade
234、inICTgoodspercapitaTimespentdealingwithgovernmentregulationsR&DspendingGraduatesinSTEMFemaleSTEMgraduatesMobile-cellulartelephonesubscriptionsInternetusers(%ofhouseholds)GendergapininternetaccessAlthough this leads to somecountrieshavingvalues5ormoreyearsoutofdate,wefeltitwasbetterforthesecountriest
235、ohaveanoldvalueratherthannovalueatall.DataAvailabilityThisyear,wewereabletoincludemorecountriesthaneverbeforeinourrankingsduetoalmosteverycountryhavingdataformorethan50%ofourindicators.Whilewewelcomethisincreaseinoveralldataavailabilityandcoveragefromourindicatorsources,therearenowanumberofcountries
236、includedintheIndexwhosescoresrelyuponmorethan40%imputeddataor,inthecaseofthe11countrieslistedabove,whosescoresarecalculatedbasedonfewerdatapoints.Intheinterestoftransparency,countrieswhosescoresarebasedonmorethan40%imputeddataarelistedbelow:AndorraAntiguaandBarbudaCentralAfricanRepublicComorosDemocr
237、aticPeoplesRepublicofKoreaDominicaEquatorialGuineaEritreaGrenada45GuineaBissauLibyaLiechtensteinMarshallIslandsMonacoNauruSaintKittsandNevisSaintVincentandtheGrenadinesSanMarinoSaoTomeandPrincipeSomaliaSouthSudanSyrianArabRepublicTaiwanTongaTuvaluVanuatu46GovernmentAIReadinessIndex2023Annex II:Detai
238、led scoresGlobalRankingCountryTotalscoreGovernmentPillarTechnologySectorPillarData&InfrastructurePillar1UnitedStatesofAmerica84.8086.0481.0287.322Singapore81.9790.4066.1989.323UnitedKingdom78.5782.5068.8084.424Finland77.3788.3460.3683.395Canada77.0785.3064.7381.176France76.0784.0360.4083.807Republic
239、ofKorea75.6587.5554.3685.028Germany75.2680.7863.2881.729Japan75.0882.7656.8585.6110Netherlands74.4778.9061.9682.5511Denmark73.9184.1159.9877.6512Australia73.8983.3452.5785.7513Norway72.7181.7754.9781.4014Sweden72.5574.7062.7180.2615Austria72.3777.6956.4382.9816China70.9477.3260.7674.7517Estonia70.86
240、80.5452.5279.5418UnitedArabEmirates70.4278.3256.6776.2819Taiwan70.2575.3354.5880.8520Ireland69.8271.5156.9681.0021Iceland69.5976.2152.7879.8022Luxembourg69.4183.1146.5178.6023Malaysia68.7179.9954.1372.0024Switzerland68.5757.3362.9685.4225Portugal68.2880.4850.9573.4226Italy67.6376.6150.9875.294727Spa
241、in67.4772.8650.9678.6028Belgium67.2873.0956.0272.7429SaudiArabia67.0478.7149.5972.8330Israel65.4653.6866.2976.4231Czechia65.1772.2547.7275.5532Brazil63.7072.4445.0873.5733Malta63.6480.7440.8969.3134Qatar63.5969.6444.3176.8135Lithuania63.3375.3143.7070.9936Poland63.1069.7946.8472.6637Thailand63.0377.
242、2141.3370.5538RussianFederation62.9274.1343.3871.2639Slovenia62.6371.7541.8674.2940India62.5875.1849.3963.1741Chile61.9574.5640.9170.3842Indonesia61.0376.2443.4863.3843Cyprus60.8469.3942.0471.0944Slovakia60.7367.7040.6073.9045Hungary60.6669.9642.2069.8246Uruguay60.5774.4035.3271.9947Trkiye60.5175.08
243、42.3264.1348Latvia60.3072.0738.5770.2749NewZealand60.1851.8547.0581.6650Oman58.9469.3537.7169.7651Bulgaria58.6466.0438.1771.7352Greece57.9555.9248.3769.5653Colombia57.8574.9835.2863.3054Argentina57.7270.3135.2767.5955Jordan56.8567.5640.6262.3656Bahrain56.1357.9639.2471.1957Serbia55.5774.2937.1355.30
244、58Peru54.8770.1532.2962.1848GovernmentAIReadinessIndex202359VietNam54.4869.0437.8256.5860Ukraine53.2968.9336.1854.7561Mauritius53.2769.8230.6359.3562Egypt52.6968.1940.1149.7763Liechtenstein52.6341.3847.7868.7564Romania52.3251.4239.2366.3065Philippines51.9865.4334.3856.1366DominicanRepublic50.7168.07
245、25.3458.7367Monaco50.5041.7539.9769.8068Mexico50.3743.0839.5568.4969Kuwait49.8638.1440.5870.8570Croatia49.3442.2539.3566.4271CostaRica49.1253.4133.2060.7672Kazakhstan48.5648.5630.9766.1373Azerbaijan48.1555.8630.7757.8274BruneiDarussalam48.1039.5741.8662.8775SanMarino48.0837.1144.2162.9476Lebanon47.6
246、250.5636.9955.3177SouthAfrica47.2837.8240.2263.7978Montenegro47.1540.2135.5365.7079Nauru46.7538.1938.5463.5080Andorra46.2538.8235.0464.9081Tunisia46.0748.3138.4751.4482Bangladesh46.0457.9626.0754.1083NorthMacedonia45.4039.5533.8962.7784Rwanda45.3967.8227.8740.4885Armenia45.2243.5033.5858.5786Panama4
247、3.9138.3333.2660.1587Uzbekistan43.7949.0724.8057.4988Morocco43.3437.5435.6956.7989Albania43.2642.5331.9755.284990RepublicofMoldova42.9742.6928.2158.0291Senegal42.5859.6527.6740.4392Pakistan42.2043.0034.0749.5393Bahamas42.1434.4931.4360.5194Iran(IslamicRepublicof)42.0731.5638.7755.8895SriLanka41.8942
248、.0533.0250.6096Fiji41.5737.7835.1251.8097Benin41.3761.3824.4638.2798Jamaica41.3248.0128.4147.5399Georgia41.2741.9630.3351.50100Ecuador40.8436.1327.8258.57101Kenya40.1940.0328.9551.58102Barbados40.1333.5332.8754.00103Nigeria39.8844.9424.4950.21104TrinidadandTobago39.4432.7032.2553.38105AntiguaandBarb
249、uda39.4133.3327.5657.34106SaintKittsandNevis39.2331.7128.0957.90107Belarus39.2030.9532.2854.38108Tonga39.0135.3733.5248.14109Mongolia38.9935.8327.7153.43110Botswana38.8434.8229.2052.50111Tajikistan38.7853.1820.4142.77112Seychelles38.1136.4419.5058.37113Tuvalu37.4529.2734.0149.08114Bhutan36.8836.8124
250、.3149.52115Paraguay36.8537.4524.3248.79116Guyana36.7930.6227.1952.58117BosniaandHerzegovina36.4927.9231.6849.86118SaintLucia36.4630.3727.9551.08119CaboVerde36.3036.6827.2444.98120Algeria35.9930.1030.5647.30121Guatemala35.7629.5524.9452.7850GovernmentAIReadinessIndex2023122MarshallIslands35.7030.2532
251、.2644.60123Cuba35.5235.3227.0944.15124Suriname35.5228.2726.8551.43125Namibia35.3732.0228.3045.80126Bolivia(PlurinationalStateof)35.2528.4726.6650.63127Dominica34.8227.9724.5151.97128Grenada34.6327.7725.4850.65129Belize34.2423.4429.0950.19130SaintVincentandtheGrenadines34.1330.5324.4047.47131Kyrgyzst
252、an34.1033.5322.8645.90132Uganda33.4438.0620.9641.30133Iraq33.4028.8629.0742.26134StateofPalestine33.1421.1531.9046.37135Gabon33.0925.1227.6746.47136LaoPeoplesDemocraticRepublic33.0528.3925.8144.96137UnitedRepublicofTanzania 32.8635.4620.7142.42138CteDIvoire32.7834.3320.2843.73139Honduras32.6325.7530
253、.1442.01140Ethiopia32.5937.2920.3540.11141Ghana32.5834.6623.2339.87142ElSalvador32.5823.7026.3147.72143Zambia32.3632.3220.3344.42144Vanuatu31.9126.1925.2344.32145Cambodia31.8827.9322.5345.17146Maldives31.7131.7120.7942.63147Samoa31.6627.5222.8544.61148Turkmenistan31.1720.0530.5142.94149Myanmar30.912
254、1.6327.9843.14150Nepal30.7731.0424.2137.06151Zimbabwe30.7124.0823.4144.6351152PapuaNewGuinea30.4125.8222.5642.84153Cameroon30.2730.9122.0737.84154Gambia(RepublicofThe)30.2525.1320.8144.81155Djibouti29.9519.4329.5040.92156Nicaragua29.7724.5326.3338.45157Timor-Leste29.7724.2521.3243.75158Togo29.6929.7
255、419.9639.36159Venezuela,BolivarianRepublicof29.2017.1726.1444.29160Angola29.1422.5518.5146.37161SolomonIslands29.0923.7223.4140.14162Madagascar28.4725.0119.2841.14163EquatorialGuinea28.0419.9426.1937.99164BurkinaFaso27.5027.0418.2037.26165Mali27.4524.7119.1238.52166Guinea27.4426.6519.9935.69167Kirib
256、ati27.4025.7416.6439.83168Mauritania27.0922.2122.2736.80169Eswatini27.0523.8018.2139.15170SaoTomeandPrincipe26.8625.4119.9635.22171Lesotho26.2125.4120.4632.75172Mozambique25.6221.9017.8237.15173Libya25.3110.0228.1037.81174Malawi24.8722.0020.4932.11175SierraLeone24.8321.8518.1034.52176Niger24.7125.48
257、15.8832.77177Sudan24.5115.4223.3534.75178Congo24.1922.5019.6830.40179GuineaBissau24.1115.3118.8138.22180Chad23.4421.1016.6532.56181Comoros22.6215.8720.6631.33182Liberia22.2419.2819.9027.5552GovernmentAIReadinessIndex2023183Somalia21.9818.0119.3528.58184Haiti21.978.9222.8934.10185DemocraticRepublicof
258、theCongo21.4718.3616.3929.66186Afghanistan21.2713.9921.7128.11187Burundi20.8718.4416.2127.96188Yemen19.8919.5930.949.13189CentralAfricanRepublic19.7413.5117.6428.06190Eritrea19.629.7319.4929.65191SouthSudan18.2613.3617.8023.61192SyrianArabRepublic18.1213.6728.1312.55193DemocraticPeoplesRepublicofKorea9.208.0314.275.2953